Museums are under growing pressure to do more with less.

Collections need fresh interpretation. Exhibitions need supporting content. Visitors increasingly expect multilingual access, inclusive formats and digital experiences that feel easy to use. At the same time, many museum teams are working with limited budgets, stretched staff and less time for content production.

It is easy to see why artificial intelligence is attracting attention. AI can help draft interpretation, adapt text for different audiences, support translation and reduce repetitive writing tasks. Used well, it can be a valuable assistant for curators, learning teams and digital teams. But speed alone is not enough.

In a museum context, interpretation is not just content. It carries authority. It shapes how visitors understand objects, people, places, histories and cultures. That means AI-generated interpretation cannot simply be treated as a quick copywriting shortcut.

The real question is not whether museums should use AI. The real question is: how can museums use AI while protecting trust, accuracy and curatorial accountability?

AI should assist curators, not replace them

There is a significant difference between using AI to support interpretation and handing interpretation over to an unaccountable system.

A responsible approach keeps curators, educators and subject specialists firmly in control. AI can help create a first draft, suggest alternative wording, produce summaries, support multilingual versions or adapt content for different visitor groups. But the final interpretation still needs human judgement, and that judgement matters.

Curators understand context, nuance and sensitivity. They know where uncertainty exists, which stories require care and where language may oversimplify or distort complex histories. For museums, AI should sit within a controlled workflow. It should support expertise, not replace it.

The risk is ungoverned AI

Many concerns around AI in museums are really concerns about control: control over evidence, accuracy, tone, intellectual property, cultural sensitivity and visitor trust, those concerns are valid.

If AI tools are used without clear boundaries, museums risk producing interpretation that sounds confident but lacks evidence. Translations may miss cultural meaning. Sensitive subjects may be flattened. Different versions of content may appear across platforms with no clear record of what changed or who approved it.

Visitors trust museums because they expect interpretation to have been considered, checked and approved. If AI is introduced into that process, the same standards still need to apply.

Museums should be asking:

• What source material is the AI using?
• Who reviews the output?
• How are edits tracked?
• Who signs off the final version?
• How are translations checked?
• What visitor data is collected, and why?

These questions are not barriers to innovation. They are the foundations of responsible innovation.

A better model: approved sources, review and version control

AI becomes more useful when it works from approved museum knowledge. Rather than asking AI to “write a tour” from scratch, museums can use existing curator notes, object records, exhibition scripts, education resources and approved interpretation as the foundation.

From there, AI can help transform one approved source into multiple useful formats, the value is not that AI invents the interpretation, the value is that it helps museums repurpose approved knowledge more efficiently.

This is where version control is essential. Museums need to know what changed, when it changed and who approved it, especially when content is being used across exhibitions, departments, languages or multiple venues.

The best AI workflows are not just fast. They make quality easier to manage at scale.

Multilingual and accessible interpretation

One of the strongest use cases for AI-assisted interpretation is multilingual content.

Many museums want to improve access for international visitors but cannot always afford to produce extensive interpretation in multiple languages. AI can help create first-draft translations or language variants more quickly, allowing teams to focus time on review and refinement rather than starting from scratch each time.

But human review still matters. Literal translation is not always enough. Tone, cultural context and historical sensitivity may need careful adjustment.

AI can also help museums create different levels of content for different visitor needs: shorter summaries, plain-English versions, family-friendly explanations, audio-first scripts and pre-visit information.

Combined with bring-your-own-device delivery, this can make interpretation more flexible and accessible. Visitors can use their own phones, headphones, browsers and accessibility settings, reducing reliance on dedicated hardware while creating a more familiar experience.

Visitor insight without visitor surveillance

As museums adopt more digital tools, they also need to think carefully about data. There is clear value in understanding how visitors use digital interpretation. Museums may want to know which stops are most popular, which languages are used, where visitors drop off or which routes support deeper engagement.

That insight can improve exhibitions, learning resources, staffing, marketing and tourism planning, but visitor analytics should be proportionate and privacy conscious.

Museums do not need to identify individual visitors to make better decisions. Aggregated insight can often show what is useful without collecting unnecessary personal information.

The future of museum technology should not be about gathering as much data as possible. It should be about collecting the right level of insight, for a clear purpose, with transparency and restraint.

Speed matters. Trust matters more.

AI will not replace curatorial expertise, community consultation, historical sensitivity or creative storytelling, but it can help museums work more efficiently. It can help teams create more content variants, support more languages, refresh interpretation more often and respond to different visitor needs. Used responsibly, it can free staff from repetitive drafting tasks so they can focus on quality, judgement and engagement.

The museums that benefit most will not be the ones that use AI the most aggressively, they will be the ones that use it most thoughtfully. AI can help museums interpret collections faster.

But speed is not the same as trust. And in museums, trust is the experience.

How HeriTech Media can help

HeriTech Media’s EasyGuide and SmartGuide solutions are designed to support flexible, responsible and accessible digital interpretation.

Through browser-based visitor experiences, AI-assisted content workflows, multilingual delivery, curator review processes and privacy-first analytics, museums can create richer interpretation without losing control of the story.

Whether you are planning a new exhibition, refreshing existing content, improving accessibility or exploring responsible AI for interpretation, the key is to build a process that supports both efficiency and trust.

How is your organisation balancing faster content production with curatorial accountability?